Lecture #24 ggplot II

ggplot (continued)

Preliminaries

library(ggplot2)
library(ggthemes)
library(patchwork)
library(TeachingDemos)
char2seed("Sienna")
d <- mpg # use built in mpg data frame

Plotting multiple panel graphs with patchwork

g1 <- ggplot(data=d, mapping=aes(x=displ,y=cty)) + 
  geom_point() + 
  geom_smooth()
print(g1)

g2 <- ggplot(data=d,
             mapping=aes(x=fl,fill=I("tomato"),color=I("black"))) +              geom_bar(stat="count") + 
             theme(legend.position="none")
print(g2)

g3 <- ggplot(data=d,
             mapping=aes(x=displ,fill=I("royalblue"),color=I("black"))) + 
  geom_histogram()
 print(g3)

  g4 <- ggplot(data=d, 
               mapping=aes(x=fl,y=cty,fill=fl)) + 
    geom_boxplot() + 
             theme(legend.position="none")
  
  print(g4)

# place two plots horizontally
  g1 + g2

# place 3 plots vertically
  g1 + g2 + g3 + plot_layout(ncol=1)

# change relative area of each plot
  g1 + g2 + plot_layout(ncol=1,heights=c(2,1))

  g1 + g2 + plot_layout(ncol=2,widths=c(1,2))

# add a spacer plot (under construction)
  g1 + plot_spacer() + g2

# use nested layouts
  g1 + {
    g2 + {
      g3 +
        g4 +
        plot_layout(ncol=1)
    }
  } +
    plot_layout(ncol=1)

# - operator for subtrack placement
  g1 + g2 - g3 + plot_layout(ncol=1)

# / and | for intuitive layouts
  (g1 | g2 | g3)/g4

  (g1 | g2)/(g3 | g4)

  # Add title, etc. to a patchwork
g1 + g2 + plot_annotation('This is a title', caption = 'made with patchwork')

# Change styling of patchwork elements
g1 + g2 +
  plot_annotation(
    title = 'This is a title',
    caption = 'made with patchwork',
    theme = theme(plot.title = element_text(size = 16))
  )

# Add tags to plots
g1 / (g2 | g3) +
  plot_annotation(tag_levels = 'A')

Swapping axis orientation for multipanel plot

g3a <- g3 + scale_x_reverse()
g3b <- g3 + scale_y_reverse() 
g3c <- g3 + scale_x_reverse() + scale_y_reverse()

(g3 | g3a)/(g3b | g3c)

(g3 + coord_flip() | g3a + coord_flip())/(g3b + coord_flip() | g3c + coord_flip())

ggsave(filename="MyPlot.pdf",plot=g3, device="pdf",width=20,height=20,units="cm",dpi=300)

Basic mapping of aesthetic variables

# mapping of a discrete variable to point color
m1 <- ggplot(data=mpg, mapping=(aes(x=displ,y=cty,color=class))) + 
  geom_point(size=3) 
print(m1)

# mapping of a discrete variable to point shape (<= 6)
m1 <- ggplot(data=mpg, mapping=(aes(x=displ,y=cty,shape=class))) + 
  geom_point(size=3) 
print(m1)

## Warning: The shape palette can deal with a maximum of 6 discrete values because
## more than 6 becomes difficult to discriminate; you have 7. Consider
## specifying shapes manually if you must have them.

## Warning: Removed 62 rows containing missing values (geom_point).

# mapping of a discrete variable to point size (not advised)
m1 <- ggplot(data=mpg, mapping=(aes(x=displ,y=cty,size=class))) + 
  geom_point() 
print(m1)

## Warning: Using size for a discrete variable is not advised.

# mapping a continuous variable to point size 
m1 <- ggplot(data=mpg, mapping=(aes(x=displ,y=cty,size=hwy))) + 
  geom_point() 
print(m1)

# mapping a continuous variable to point color
m1 <- ggplot(data=mpg, mapping=(aes(x=displ,y=cty,color=hwy))) + 
  geom_point(size=5) 
print(m1)

# mapping two variables to different aesthetics
m1 <- ggplot(data=mpg, mapping=(aes(x=displ,y=cty,shape=class,color=hwy))) + 
  geom_point(size=5) 
print(m1)

## Warning: The shape palette can deal with a maximum of 6 discrete values because
## more than 6 becomes difficult to discriminate; you have 7. Consider
## specifying shapes manually if you must have them.

## Warning: Removed 62 rows containing missing values (geom_point).

# use shape for smaller number of categories
m1 <- ggplot(data=mpg, mapping=(aes(x=displ,y=cty,shape=drv,color=fl))) + 
  geom_point(size=5) 

# use all 3 (size, shape, color) to indicate 5 attributes!
m1 <- ggplot(data=mpg, mapping=(aes(x=displ,y=cty,shape=drv,color=fl,size=hwy))) + 
 geom_point()
print(m1)

# mapping a variable to the same aesthetic in two different geoms
m1 <- ggplot(data=mpg, mapping=(aes(x=displ,y=cty,color=drv))) + 
  geom_point(size=2) + geom_smooth(method="lm")
print(m1)

Faceting variables for more effective groupings

# basic faceting with variables split by row, column, or both
m1 <- ggplot(data=mpg, mapping=(aes(x=displ,y=cty))) + 
  geom_point() 
m1 +  facet_grid(class~fl)     

m1 + facet_grid(class~fl, scales="free_y")

m1 + facet_grid(class~fl, scales="free")

m1 + facet_grid(.~class)

m1 + facet_grid(class~.)

# use facet wrap when variables are not crossed
m1 + facet_grid(.~class)

m1 +  facet_wrap(~class)     

m1 + facet_wrap(~class + fl)

m1 + facet_wrap(~class + fl, drop=FALSE)

m1 + facet_grid(class~fl)

# use facet with other aesthetic mapping within rows or columns

m1 <- ggplot(data=mpg, mapping=(aes(x=displ,y=cty,color=drv))) + 
  geom_point() 
m1 + facet_grid(.~class)

# easy to switch to other geoms
m1 <- ggplot(data=mpg, mapping=(aes(x=displ,y=cty,color=drv))) + 
  geom_smooth(se=FALSE,method="lm") 
m1 + facet_grid(.~class)

# fitting with boxplots over a continuous variable
m1 <- ggplot(data=mpg, mapping=(aes(x=displ,y=cty))) + 
  geom_boxplot()
m1 + facet_grid(.~class)

# add a group and fill mapping for subgroups
m1 <- ggplot(data=mpg, mapping=(aes(x=displ,y=cty,group=drv,fill=drv))) + 
  geom_boxplot()
m1 + facet_grid(.~class)

d <- mpg # use built in mpg data frame

Mapping aesthetics within geoms

# standard plot with all data
p1 <- ggplot(data=d,mapping=aes(x=displ,y=hwy)) +
  geom_point() + geom_smooth()
print(p1)

# break out the drive types (note what group affects
p1 <- ggplot(data=d,mapping=aes(x=displ,y=hwy, group=drv)) +
  geom_point() + geom_smooth()
print(p1)

# break out the drive types (note what color affects
p1 <- ggplot(data=d,mapping=aes(x=displ,y=hwy, color=drv)) +
  geom_point() + geom_smooth()
print(p1)

# break out the drive types (note what fill affects
p1 <- ggplot(data=d,mapping=aes(x=displ,y=hwy, fill=drv)) +
  geom_point() + geom_smooth()
print(p1)

# use both if you want points, lines, and confidence intervals colored
p1 <- ggplot(data=d,mapping=aes(x=displ,y=hwy, color=drv, fill=drv)) +
  geom_point() + geom_smooth()
print(p1)

# now use aesthetic mappings within each geom to over-ride defaults
# subset the data frame to pull out what you need

p1 <- ggplot(data=d,mapping=aes(x=displ,y=hwy,col=drv)) +
  geom_point(data=d[d$drv=="4",]) + geom_smooth()
print(p1)

# instead of subsetting, just map an aesthetic
p1 <- ggplot(data=d,mapping=aes(x=displ,y=hwy)) +
  geom_point(mapping=aes(color=drv)) + geom_smooth()
print(p1)

# Conversely, map the smoother, but not the points
p1 <- ggplot(data=d,mapping=aes(x=displ,y=hwy)) +
  geom_point() + geom_smooth(mapping=aes(color=drv))
print(p1)

# also, subset in the first layer to eliminate some data entirely
# instead of subsetting, just map an aesthetic
p1 <- ggplot(data=d[d$drv!="4",],mapping=aes(x=displ,y=hwy)) +
  geom_point(mapping=aes(color=drv)) + geom_smooth()
print(p1)